Research in AI for Protein Design @Harvard | Prev. CS PhD @UniofOxford, Maths & Physics @Polytechnique
Boston
Joined September 2020
- Tweets450
- Following407
- Followers1.4K
- Likes1.9K
Yes based on our paper, but setup is not very clear wrt how models are prompted, whether zero-shot or few shot, whether papers are in training set, etc
Pascal Notin retweeted
Binder design is nice and all, but here we have three agents sharing data and autonomously controlling a lab to design enzymes with shifted substrate scopes and high activity!
@cobanbrooks @NotinPascal @romerolab1
Pascal Notin retweeted
I’m really excited to share our new preprint, out today, where we built a fully autonomous system for enzyme engineering — integrating a self-driving lab with a generative protein language model that worked together over a month to engineer substrate specificity in glycoside hydrolase (GH1) enzymes.
biorxiv.org/content/10.64898…
[1/n]
It most likely is based on our work. But what the "hard subset" is or how the LLMs are actually evaluated on it is not disclosed anywhere afaik. Not to mention all assays in PG are based on published work / could have landed in the LLM training data
Thanks @rlacombe. I must say I have no idea what the "ProteinGym hard" benchmark used in these system cards is. I looked for a description somewhere but could not find anything. If anybody knows I'm interested
Pascal Notin retweeted
Now available as its peer-reviewed version in Nature:
nature.com/articles/s41586-0…
Thrilled to share our new paper where we introduce a multiplexed hydrogen–deuterium exchange MS (mHDX‑MS) method that can measure hundreds of protein domains’ conformational energy landscapes—all in a single experiment!
biorxiv.org/content/10.1101/…
Pascal Notin retweeted
Quite pleased to hear that one of my submitted proteins placed #5 out of 1200 that were tested in this lovely competition. The approach: pure rational design. Nice to see that human + Microsoft Word is still competitive with state-of-the-art AI methods 🙂
The results of the Nipah Protein Design Competition are out!
🧬 1200 proteins experimentally validated (3x more than last year)
📈 99 novel binders against the target protein (a challenging tetramer with little prior work)
💪 26 single digit nM or better binders, with the best ones at single-digit picomolar affinity!
All data now available open-source on Proteinbase!
Let's take a look at the results ⬇️
Pascal Notin retweeted
Super excited to share that I will join @Princeton @omenndarlingbio this Fall as an assistant professor. We combine syn bio and AI to program macromolecules and cells for therapeutics engineering and drug target discovery. jianglab.bioengineering.prin…
Pascal Notin retweeted
New paper “Proteome-wide model for human disease genetics” is now live at Nature Genetics: rdcu.be/eRu7K
popEVE (pop.evemodel.org) finds the needles in the haystacks of human genetic variation:
You both may find this work interesting -- better performance than all the aforementioned models on ProteinGym, and retrieval 2 orders of magnitude faster vs mmseq-GPU: proceedings.mlr.press/v267/w…
Pascal Notin retweeted
Reminder - PhD applications for OATML are now open
The first funding deadline is December 2 - candidates interested in developing Bayesian deep learning methodology, applications of ML, AI security, and understanding ML methodology are encouraged to apply
More info: oatml.cs.ox.ac.uk/apply.html
Pascal Notin retweeted
Thrilled to announce I'm starting as a Principal Investigator at #Aithyra in Vienna! We'll be developing generative models to understand cell biology and design proteins.
I'm hiring PhDs, Postdocs, & Visiting Researchers! PhD applications by Sept 10: apply.cemm.at
Pascal Notin retweeted
Our paper on generalizable antibody-antigen binding affinity prediction has been featured on the cover of the August Issue of @NatComputSci! 📔🎉
🚨Our August issue is now live and includes research on antibody-antigen binding, molecular screening for zeolite synthesis, psychological experiments with LLMs, and much more!
nature.com/natcomputsci/volu…
ALT Yellow and orange antibody binding to purple and blue membrane proteins
Congrats @houchao1 ! Wondering if NLL -- as a measure of how a given protein is constrained -- could be used as a quick diagnostic to assess which protein is "easier to evolve" in protein engineering campaigns
Congrats @BorisMPower and team! Would love to test out the approach on DMS assays in ProteinGym (proteingym.org). Do reach out if interested!
Pascal Notin retweeted
I’ve thoroughly enjoyed reading two (VERY!) recent papers that model protein sequences by retrieving evolutionary information (dynamically) at inference time, and there's a lot to unpack!
[1] arxiv.org/abs/2506.08954
[2] biorxiv.org/content/10.1101/…
(1/n)
Pascal Notin retweeted
A generative artificial-intelligence tool has designed a synthetic CRISPR system that successfully edits human DNA
go.nature.com/3JadkRy
Pascal Notin retweeted
1/5 Biological data is noisy, redundant, and ever-growing. 🗣️
In our new paper (first paper of my post doc!! ⚡️), we track model performance across 14 years of UniRef100 snapshots to ask: how does pLM performance scale with training data?
